Introduction To The Study Of The History Of Language

x.] ON ISOLATION AND REACTION AGAINST IT. 183 indefinite article. Few are unaware that when un stands before a consonant the n is not pronounced, leaving in the spoken word only a trace of its existence in the fact that the vowel is nasalised. When un comes before a vowel, on the other hand, the vowel is much less strongly, if at all, nasalised, and the n is clearly pronounced. Thus (using the circumflex to indicate the nasal quality of the vowel and o for the sound of =u in un), un pere o pere, but un ami= on ami or on ami. The corresponding difference which exists in English is expressed in writing : a father, an aunt. Just as the article is closely connected with the noun, so preposition and noun, or preposition and verb, are very intimately connected in pronunciation. Hence though many, who have never carefully ob- served either their own pronunciation or that of others, may dispute or deny the assertion in ORDINARY con- versation, in the phrases, in town, in doors, we employ the n sound but when the word in stands before Paris ; mand Berlin, we use an sound, just as we say impos- sible by the side of interest. Similarly, we pronounce generally 'in coming' with ng for n, just as we speak of a man's ingcome. This differentiation of the pronun- ciation of the preposition in into three forms in, im, ing is not, however, consistently expressed by us in writing. The Greeks, on the other hand, who similarly differentiated the terminal consonants of the preposi- tions in their spoken language, but on a much larger scale (accustomed as they were to a far closer correspondence between their spoken and their written language than the Englishman observes), did actually, in many cases, write as they spoke : /ca Se, KCLK /ce^aX^i/, Kayyow KOLTT TT&IOV, etc., instead of employing the nor- mal form of the preposition, Kara. So we find in inscrip- tions T7]p, TroKw, TJ)y yvvaiKa, roX Xoyd^, e/x, TrdXet, etc.

184 THE HISTORY OF LANGUAGE. [CHAP. The first step on the road towards unification is frequently that the external reason which caused the difference in form, disappears or loses force, and one form is found in connections where, historically or Wephonetically speaking, the other is correct. may instance this by the common mistake of children when they say, e.g., a apple instead of an apple. In this case, however, the correct form is so very frequently heard that the encroachment of a on the domains of an is not likely to lead to permanent confusion. Where, how- ever, circumstances are less favourable to the preserva- tion of the historically correct usage, it happens that either form encroaches on the domain of the other, or else it may result that the encroachment is reciprocal, when, after a period of confusion in which both forms are used indifferently, one becomes obsolete and falls into oblivion, not without often leaving some striking form or phrase to testify to what once existed. Thus, for instance, our word here, Old High German hier, or h$r, was, in the period of transition from Old to Middle High German, differentiated in accordance with a phonetic law of that time, viz. that final r was dropped after a long vowel. If not final however, r remained untouched, and this whether it stood in the body of a word or within a group of intimately con- nected words. Of the two forms hie and hier, the former, as the form employed when the word was used independently, was in Middle High German often set before words beginning with a vowel and we find hie ; inne = or even, by contraction, hinne, for hier- ( here-in) inne. On the other hand, it is probably owing to the frequency of combinations similar and equivalent to our here-in, here-upon, etc., that the form hier encroached successfully upon the domain of hie, and finally supplanted it. Hie, however, remained, singu-

x.] ON ISOLATION AND REACTION AGAINST IT. 185 larly enough, in the one expression hie und da (here and there), where the form without r is not and Anhas never been, phonetically speaking, correct. excellent example of this differentiation is furnished by one, an. The best example of the process is furnished by the history of the working of Verner's law, and of thegradual Wedisappearance of its effects. have before (pp. 172, 1 73) explained this law and quoted instances of forms created in agreement with it, which have now been replaced by others. To repeat this explanation here with other examples would be superfluous ; to give a full history, even confining ourselves to an enumera- tion of all the various ways in which it has been opera- tive and the areas of its influence, would transcend the scope of this work. To carefully note all instances of its occurrence and its neglect, and to closely investigate the possible courses of the latter, is a task which may most usefully challenge the attention of philologists. We will illustrate the truth of this by a single example: (though even this we cannot discuss exhaustively). The forms which we employ at present as the past tense of the verb to be sing, was and plur. (with grammatical change according to the law) were, belong to a root which in old English and Anglo-Saxon furnished a complete verb : pres. wese, past, wees, Nowp. part, wesen. we should naturally expect that in a time when the grammatical change was still pre- served in fredse, frdas, fruron, froren, (to freeze) etc. ceose cdas, curon, coren> (to choose) y seffie, sed$, sudon, soden, (to seethe, to boil) we should also find that change here, and that accord- ingly the past participle should be *weren. That such

1 86 THE HISTORY OF LANGUAGE. [CHAP. a form once existed is proved by the past participle forweorone (cf. Sievers, Anglo-Saxon Grammar, 391). Everywhere, however, in Anglo-Saxon, in the past participle of this verb and in that of all similarly con- jugated, such as lesan, Ices, lesen ; genesan, gences, gene- sen, etc., the s has once more been fully established. The fact that these past participles had already so far proceeded on the road to unification, while the others as yet remained isolated, may be explained in this way, the latter, IN ADDITION to the differentiation in accordance with Verner's law, showed a difference of vowel-sound, which in the case of others did not exist. Hence the forms differentiated in two distinct ways were able to resist the tendency towards unification long after those which differed only in one respect had succumbed. In fact, of the former we still have such remnants as forlorn, from to lose ; sodden, from to seethe. We may formulate the result which we have illustrated, thus : The greater the phonetic distance of two differenti- atedforms, the greater is the power of resistance against unification and equalisation. But the ORDER in which we see the traces of the working of Verner's law disappear one after another, and the study of such few remnants as still exist, brings out two other general truths concerning unification. We may without hesitation affirm that, close as is the etymological connection between the various tenses of the same verb, or, to speak perhaps more correctly, that clearly as that connection is felt by the speech-making community, it is still more strongly felt as between the various forms of the same tense, or the various cases of the same noun. Now, it is against the differentia- tion between the members of these most intimate groups that unification first takes place. In the declen- sion of the noun, where nothing but the operation of

x.] ON ISOLATION AND REACTION AGAINST IT. 187 Verner's law had separated the various cases, the re- assimilation first took place, and though we can prove that, in this case also, the differences actually once existed in the historic periods of the Teutonic dialects almost all traces thereof have been obliterated. In the past tenses of the verbs they are still at first found, supported as the differentiation had been by that other force the gradation of vowels (the 1 But 'ablaut'). again : unification between the singular and plural of the past tense took place first in cases where the vowels were alike in both, and next in those where the vowels differed and again, this occurred before the unification of the past participle with the whole group. In agreement with this same rule, that very difference of vowel-sound has completely disappeared in all past singulars and plurals, even where as, e.g., in German generally the past participle still preserves the ' ablaut.' We can then lay it down as a second rule, that the closer the etymological connection is between dif- ferentiatedforms-, the sooner will unification be effected ; whilst a consideration of such rare instances as the preservation of the interchange of s and r in / was, we were, which is clearly due to the very exceptional fre- quency with which these forms must always have been used, and the consequent firmness with which they are impressed on every speaker's memory, exhibits a third law, viz. that the greater the intensity with which differ- entiatedforms are impressed upon the minds of the com- munity, the greater will prove their power of resistance against unification. It is further evident that in cases where the dif- 1 And by the expectation thus created of the regular occurrence of such differentiation between past singular and past plural, even where this ablaut did not show different vowels.

1 88 THE HISTORY OF LANGUAGE. [CHAP. ferentiation of form had been accompanied by one in meaning, the tendency towards unification was counter- acted, or rather can never have existed. Thus, the =pair of words glass (etymologically the shining sub- stance) and glare (to shine) is separated once and for Weever. have seen the plur. dawes re-united to sing. day ; the verb to dawn has not followed suit. Though thus much is clear, and when once appre- hended, almost self-evident, we must acknowledge that much is as yet obscure and unexplained. It is often already very difficult to find any reason why in one case unification has taken place and not in another, which apparently presented the same conditions : it is generally harder still to find an answer to the ques- tion why in any given case one form has prevailed over another, instead of the converse having happened. Omniscience alone could answer all such questions : but here, again, a few general observations may serve to explain some points, though, as we have said, much as yet remains inexplicable. Thus, for example, when unification replaces the confusion which followed differ- entiation, members of the same formal or modal group (that is to say, for instance, the same parts of speech) are likely to follow in the same direction. Thus, e.g., in the original Teutonic, when the suffix no was pre- ceded by a vowel, that vowel varied in the different (strong and weak) cases of the declensions of nouns, adjectives, and participles, according to fixed rules, between u and e. This u developed into o or a, and e into i. Soon unification took place, in some cases in one, in others in another direction, so that we find, for instance, in Gothic a form like ftiudAns (king) by the side of matirgms (morning), whilst now, the past par- ticiples (formed with this same suffix) all have ans throughout ; such participles as became pure adjectives

x.] ON ISOLATION AND REACTION AGAINST IT. 189 or^ nouns have often ins, e.g. gafulgins (adj. ' secret '), past participle, of filhan, ' to hide/ with fulhans as past participle, = hidden aigin (neuter, hence with- ; out s in nom.) = property, is past participle of aigan, ' to have.' Sometimes as, for instance, in the singular and plural of past tense in strong verbs a differentiation coincides with difference in function, though its origin was independent of any such functional divergence. This, of course, strengthens the phonetic differentia- tion, and, if such a coincidence affects simultaneously a formal group of large extent, and thus becomes a model for analogical formations (Chap. V.), the originally meaningless phonetic divergence may become indis- solubly associated with difference of function, and so become expressive of the latter. Thus, for instance, the words tooth, foot, and man form their plural teeth, feet, and men by umlaut, and by umlaut alone. This modification of the vowel is, then, here expressive of plurality. Originally, however, it was not so. In Anglo-Saxon the declension was Singular Nom. and Ace. f6t tfo mann t6>es Gen. fotes mannes Dat. fit t& menn menn Plur. Nom. and Ace. fa te?> manna fdta toa mannum foturn tffi&um When once the combined force of nominative, accu- sative, and genitive had ousted the modified vowel from the dative singular, the whole singular exhibited 6 (a) in contrast to the nominative and accusative plural with ^ (e). This caused the transference of the latter to the genitive and dative plural also, and thus

1 90 THE HISTORY OF LANGUAGE. [CHAP. x. invested the modification with a force originally quite foreign to it. In English, no doubt owing to the mixed influence upon that language of two very different grammatical systems (the Teutonic of Anglo-Saxon, and the Ro- mance of Norman-French), unification has proceeded to a far greater length than in most other Teutonic dia- lects. In German, e.g., the history of the ^tmlaut and the origin of plurals in er of which English has no trace but the provincialism childer, or the \"correct\" form children furnish examples of what we have said ; and students of German will find a careful investigation of that history both interesting and instructive.

CHAPTER XI. THE FORMATION OF NEW GROUPS. THE effect of sound-change is to produce differences in language where none previously existed ; but it likewise tends to cancel existing differences, and to cause forms originally distinct to resemble each other or actually to coincide. Now, symmetry and uniformity are clearly an aid to the memory, when attained by the abolition of useless and purposeless differences. It is, for instance, in English, far simpler to state, and far more easy to remember the statement, that all plurals are formed by adding s to the singular, than that some are formed in -n, or -en, or by such modifications as man, men ; foot, feet ; etc. : and it is therefore a gain to language when such forms as shoon, eyen, etc., disappear Onin favour of such forms as sfioes, eyes, etc. the other hand, the cancelling of such differences when they serve to mark different functions is naturally dis- advantageous and tends to obscurity. When a sound which marked such a functional difference disappears, or when of two words or forms which had different meanings one becomes obsolete, and the other is employed to do service for both, it is clear that language cannot but be the loser by dispensing with an important aid to clearness and distinction. Thus, of the two forms mot and moste, the former has now

1 92 THE HISTORY OF LANGUAGE. [CHAP. disappeared, and the latter, in the form must, serves to indicate both the present and the past tense. The effect of this ambiguity is that where we wish to clearly indicate the past of must, we have to employ some idiom in which must has no place ; as ' was obliged to/ 'had to/ 'was constrained to/ etc. Similarly, the loss of the plural s in very many French nouns (which s, though still written, is seldom sounded) would create ambiguity were it not that the difference of the article attached to the noun marks the difference, and to a large extent remedies the evil cf. Vami, les amis. ; The remedy, however, for such obscurity is not always to be found in the context. Sometimes, indeed, the evil brings its own cure changes arise which ; enable the necessary distinctions to be once more felt and maintained, creating new forms by analogy with other forms (see Chapter V.) : but, on the other hand, it frequently occurs that the evil remains, and a con- fusion follows in the grouping of the words which ; grouping, as we have seen, is all-important in the life history of the members of the group. We must in this chapter endeavour to study some of the results of this confusion, and consequent re- arrangement in the groups ; and to distinguish the cases where similarity caused by phonetic development affects the matter-groups from those where the modal- groups are influenced. I. i. There are many cases where words connected neither by etymology nor by signification fall into the same form. Still, in spite of this similarity in form, the words remain perfectly distinct in the linguistic conscious- ness of a speaker of ordinary intelligence. Such are, i. a. Hale, in such a phrase as hale and hearty.

xi.] THE FORMATION OF NEW GROUPS. 193 This word is of Scandinavian l origin Icelandic (cf. Aeill), and represents the Anglo-Saxon Ml, to which word we owe the misspelt word whole, b. Plate, ' to drag/ found in Middle-English as halien. 2. a. Whole =K.S. hdl ; see above, b. Hole = A.S. hoi, 'a cave.' 2 3. a. Grave (A.S. grdfan). b. Grave (Fr. grave, Lat. graveni}? 4. a. Cope (O.Fr. cape), b. Cope (Dutch koopen = to bargain, to chaffer, to buy, to vie with). 5. a. Stile (A.S. stigel). b. Stile (commonly misspelt style, Lat. stilum). 6. a. Well, adverb (A.S. wet), b. Well, noun (A.S. wella). Arm7. a. (Lat. arma). b. Arm, the limb, cog- nate with Ger. arm. 8. a. Lay (A.S. /^0*). b. Lay (O.Fr. 'song'). 9. 0. /W<? (Fr. /#/, Lat. palum). b. Pale (Fr pdle, Lat. pallidum). 10. 0. .ff/dfer, the tree (A.S. <?/&/%). b. Elder < older.' 1 Thus says Professor Skeat, Etymological Dictionary, s.v. ; others maintain that it is due to Northumbrian preservation of a, which in the South became o. 2 Professor Skeat (Principles of English Etymology, p. 411) draws a useful distinction between homographs and homophones, or words spelled alike and those sounded alike. For our purpose, as students of the spoken language, the homophones alone are of importance. homograph A is commonly, but not invariably, a homophone; cf. 'I read now' and 'Yesterday I read: We need not here further consider such vagaries of English spelling. 8 It is unnecessary to point out in the text that we must bear in mind that French nouns or adjectives are almost always derived from the accusative case as representative of the oblique cases. For the full explanation of this see Brachet's Grammaire Historique Introd.

194 THE HISTORY OF LANGUAGE. [CHAP. It would, of course, be possible to extend this list to almost any length ; but this would be useless for our purpose, which is to investigate solely those cases in which similarity causes confusion. This happens where the difference in origin and meaning is lost sight of. It is naturally impossible to draw a hard and fast line of demarcation between the case just discussed and that which we are about to exemplify, as one speaker may keep distinct what another may confuse or treat as identical. Still, no one, we may fairly say, unless he be a student of language, or unless he has been expressly informed, is aware that in a phrase like The ship is boundfor London, the word bound employed by him has absolutely no connection with the past parti- ciple of the verb to bind. In the first case, bound is of Scandinavian origin, and meant originally ready, pre- pared; cf. the Icelandic verb bua, perf. part, b&inn, 'to prepare.' Similarly, few ordinary speakers can explain, or indeed realise, the existence of the distinction in meaning between shed, ' (a doublet of shade), 'a hut and shed in water-shed, when derived from the A.S. sctadan; or that between sheer, allied to Icelandic skarr, ' and sheer, akin to Dutch scheren, ' to bright/ some shave.' Thus, again, many might suppose that ' etymological connection existed between hide, a( skin (A.S. hyd, akin to Ger. haui), and hide, 'to conceal' (A.S. hidari) ; while others, when told that hide also served as the name for a certain measure of land, might naturally even suspect some allusion to the famous legend of the foundation of Byrsa or Carthage. The A.S. noun setl (a seat) and the verb settan survive both in the word settle and in to settle. In employing, however, the word in ' to settle a dispute/ we have a word of very different origin : the A.S. sacu, a' quarrel/ 1 mydispute/ 'lawsuit' (surviving in 'for sake,' etc.),

XL] THE FORMATION OF NEW GROUPS. 195 existed side by side with a verb sacan, ' to strive/ or ' akin to this, we find saht, a substantive dispute:' which owes its meaning, ' to the de- reconciliation/ velopment lawsuit, adjustment by lawsuit, etc. Again, derived from this we have the verb sahtlian, 'to reconcile/ which, at a later period, occurs in the forms saztlen and sattle} When this verb ceased to be understood, confusion with the other verb =to settle to fix, to arrange, arose, and the two forms ' flowed together, just as two drops of rain running down a window-pane are very likely to run into one/ 2 Another instance of this nature is discussed by Professor Skeat, s.v. ; viz., sound = A.S. sund, akin to the Ger. (ge)sund; sound, 'a strait of the sea/ and sound, M.E. soun, Anglo-Fr. soun or sun, Lat. sonum. ii. Such forms, where phonetic development brought about merely a close resemblance without producing perfect similarity, and where, as a next step, one or other of the set of words underwent some change more or less violent in consequence of its supposed connec- tion with the rest, are peculiarly instructive, proving as they do the confusion which arose in the minds of the speakers who thus combined what was distinct and unconnected. In these cases we have entered upon the domain of 'popular etymology/ to which we have already incidentally alluded. It does not, however, always follow that the supposed connection in meaning in other words, the coalescence of elements of different origin into a single material group, brings about the further change in form ; at this period nothing but the linguistic con- sciousness of the speaker can decide whether the 1 See Skeat, Etymological Dictionary, s.v. settle; Stratman sv sahtlen. Skeat, Principles of English Etymology, p. 410.

196 THE HISTORY OF LANGUAGE. [CHAP. 'popular etymology' is or has been at work. Of course, as long as the etymology of the different words in the set is clearly understood by the speaker, there can be no question as to the connection, but when one or more of the members of the set is no longer under- stood in its historical bearings, it is possible for a new grouping to arise. Let us take, as an instance, the word carousal. This bore originally the sense which it bears in the Parisian name of the Place du Carrousel, viz. a tourna- ment or festival. It was confused with the word carouse (Ger. gar-aus = properly ' out,' i.e. ' quite empty your glasses ') ; and at present our word carousal represents both. The Anglo-Saxon word bonda meant a boor, or householder. His tenure appears expressed in Low Latin by the word bondagium, and it is only to a supposed, but wholly erroneous connection with bond and the verb to bind, that our present word bondage owes its sense of servitude. The Fr. sursis gave us, before its final s had ceased to be pronounced, our verb surcease, which most speakers now look on as a compound of cease (Fr. cesser}> Wiseacre, really derived through the Dutch from the Ger. wizago (A.S. witega, ' a prophet '), was already, while on its way to England, misunder- stood in Holland, and taken to be a compound of wise. In Dutch, a verb wys-seggen and a noun wys-segger to speak ' and a' wise sayer ') were formed, (' wisely 1 Or rather Fr. (je) cesse. Just as, in the French language, we must explain most nouns from the Latin accusative form, so in English most of the verbs which we owe to French can only be explained by the '' forms, e.g. first person singular of the strong present tense as complain iiomje complain, and not from complaindre; ; to despise, O.Fr. tu despis, not infinitive despire ; to prevail, je prevail, not prevaloir; to relieve, je (re)lieve, not from relever ; to acquire, facquier, not from acquerir.

XL] THE FORMATION OF NEW GROUPS. 197 and modern German as well possesses the word weissagen, 'to prophesy.' This wys-segger, when it reached England, could no longer be understood as a derivative from the verb secgan, which in English had already lost its guttural and had become (to) say ; and thus popular etymology altered the second part of the supposed compound into the meaningless acre. The Fr. surlonge, the piece of meat 'upon the loin' (Lat. super, Fr. sur, and Lat. *lumbea, from lumbus, Fr. longe), became in English the surloyn in the time of Henry VI. This was no longer understood ; the word was accepted as a compound with the word sir, and thus the fable was invented of the ' monarch ' knight- merry ing the 1 The berfroit or belefreit of Old French loin. is of German origin, and signifies a watchtower. The word had ceased to be understood, and its origin was forgotten ; but, as many towers contained a bell or a peal of bells, a supposed connection with these bells caused the word to be changed into belfry. The gspelling is affected in sovereign, where the is due to a supposed connection with to reign (re'gner, regnare) ; the real derivation being from soverain (superaneum), and the word being correctly spelt sovran by Milton. Further instances are lance-knight = lanz-knecht = ( landes knecht = ' the knight, i.e. the man~of the land' ' the servant of his ' cray-fish = country ) ; ( e'cre'visse) ; shamefaced (really shamefast, like steadfast), etc. In other cases of rarer occurrence than those which we have discussed, a significant part of a compound assumes the form of a mere derivative. This has occurred in the case of the word righteous, taken to be a derivative from some French adjective in -eux, Lat. -osus, though really due to right-wise, a compound like otherwise. It is natural that Proper nouns, where 1 See Skeat, s.v.

Tin.; there is no connection or only a lam iiul one between the word and its meaning, should be more liable to such transformations than others so the Rose des ; quatre saisons appears as the quarter-sessions rose, the asparagus appears as sparrow grass, the ship Belle- ;,</-, IM, <>mrs the /////)' rif//tlH* the /'/,'/, V.v.W, the Wetearing kisser. may perhaps add here a word lik<- liquorice, which, though the name, rightly under- stood, is dcs< TI|>U\\< -, has become a mere proper noun. Originally from liquiritia, itself a corrupt form of fflykyrrhisa ** &' sweet root/ it has, as its spell! n. shows, become connected with /it/nor? while those \\vh<> <|rmir<l ihr. impossible |>refei red to explain the woid as connected with to lick? II. I mportant, then, as the part played by phom i ic development is in I > ringing about the formation of new material-groups, it has made its influence i< li more widely still in the modal grouping of the various .systems of inlleetion. Mere, a-.im, two cases should le distinguished: (i) when forms which have had identical functions come to coincide: (2) when such coincidence occurs in the case of forms that have had dillerent him lions. The1. cancel lino of diversities in lot in or in in Heel ion when MII h inlleetion indicated no difference in function raust obviously on the whole be set down as a gain to language: simplicity is gained thereby without any loss in clearness. This gain, however, is only effected 1 It appears that this, and not Bitty ruffian^ is the form used by Milois. MVIMll \\\\.MiKI ilur. 111. il /'V//r /////.,/>/ is a luitlu-r popular etymology, due to 'schol u Src r.ilmn. |-olk Ivlvnioloi-v. s.\\ 1 This derivation is given in a certain well-known SCHOOL edition of Milton's Comus : liquorice something which makes one lich one's

XI. Till' T'oKMATION OF NEW GROIN I when (he abolition is complete; should ihe abolition be partial only, simplification may be. gained at the expense of a new confusion. We have an example of such a complete process of cancel I iiu; in the terminations er and est in the com- parative and superlative ol adject In (inline the comparative was formed either with the suffix iz or 6z, the superlative with is/ or with^v// and, except, indeed, i hat the forms in iz and ist were more common than those in 6z and 6st, and that the latter are found only with stems in a, no rule can be given for their occur- rence. Thus w<iir<igs (an a stem) has in its compara- tive managiz-a, superlative managists ; afaeis (ja stem) afaiza, afaists ; hardiis (u stem), hardiza, hardists ; but //Wis\\, fnxttza, frdddsts ; arms, armdza, armdsts. 1 In Old High German there was a similar uncertainty. Here the z of Gothic appeared as r in the 2 comparatives, and while W/V has for its comparative W/<wv and its wesuperlative saltgdsto, find (//)/r////, (li^rciniro, (/t)rei- nisto? In Anglo-Saxon we lind already but a single termination for the comparative, viz. ra; but the two forms of superlative are still extant in ostandest; earm, earmra, earmost ; //<;m/, /iranlra, heardost ; but eald iddra (with umlaut or modified vowel),* ieldest. Our forms hard, harder, hardest; old, older, oldest ; silly, sillier, silliest, etc., are clearly a further step in the rii-Ju direction of simplicity in system. The convergence is, however, not always complete : sometimes it happens that two systems coincide and ; , Goth. Gram., 135-137. I'oi Miml.ii interchanges of r and z (s), cf. Latin Venus, Veneris for * Vencsis ; arbos, arboris for *arbosis, etc. 11 P.i.imic, Alt I lorlulciitsc lie ( ii.iin., 260 Sq<|. 4 The term umlaut is more convt nn ni tli.m ' modification of the vowel sound.'

200 THE HISTORY OF LANGUAGE. [CHAP. this coincidence may be (i) in ALL FORMS but only in SOME WORDS belonging- to each system ; or, again, (2) it may manifest itself in ALL WORDS but only in SOME FORMS and, lastly, this coincidence may affect (3) only ; SOME WORDS in SOME FORMS of two converging systems. In the case of (i) the convergence is complete and irrevocable, and words which formerly belonged to one system have simply parted company with it, and have definitely joined the other to which they were assimi- lated. In. the cases, however, of (2) and (3), confusion must arise, and further development must be looked Wefor. find a good illustration of this confusion and of its development in the history of the Teutonic declensions. In the case of these, as of other Indo- European languages, the declensions differed as the stems of the words terminated in a consonant or a vowel and amongst the latter, again, we must draw ; distinctions between the declension of stems in a, (o\\ i, and u. In the a declension, again, a subdivision arose tor pure a, ja, wa, and long a stems. These different terminations of the stems are, for instance, clearly pre- served in Gothic dat. and ace. plur. dags, dagam, dagans ; gasts, gastim, gastins ; sunus, sunum, sununs ; and (with Gothic o instead of a) giba, gibom, gibos. In the oldest forms of Scandinavian, the so-called Ur-Norse, also, we find the vowels preserved in the nominative singular, holingar, erilar, ztc.,gastir, staldir, etc., hauktf&ur, warur: 1 but even in these, the oldest forms of the Teutonic dialects accessible to us, the various systems were confused ; and it is the study of Comparative Grammar that we have to thank for the distinction between the different classes and, again, it ; is only owing to the light shed on the subject by the comparison with Latin, Greek, and Sanskrit cognates, 1 Noreen, Altisl. Gram., 266, 299, 307.

XT.] THE FORMATION OF NEW GROUPS. 201 that we are enabled in some instances to decide to which of these classes any given word belongs. The ' wear- ing down ' of the various terminations produced here identity, elsewhere close resemblance of many cases in many words, while in other cases the influence of the preceding letter made itself felt, and a difference in declension arose for the a stems : this difference de- pending on whether the a was preceded by a con- sonant i (/) or w. Where phonetic development had caused some of the cases to agree, other cases soon followed suit, and thus we find, for instance, that even in Gothic the entire singular of i declension has already become identical with that of the a stems : Sing. Nom. a stem. / stem. Gen. dags balgs Dat. dagis balgis Ace. daga balga Voc. balg dag balg Plur. Nom. dag balgeis balg Gen. dagds balgim Dat. dag$ balgins. Ace. dagam dagans As a consequence of this, numerous words which cognate languages prove to belong to the i declension are nevertheless entirely declined like a stems in Gothic and even in the very few Gothic texts which ; we possess, and though these are derived from one source only, we meet with words evidencing the fact that Ulfilas himself (or, it may be, his copyist) was sometimes confused as to the declension usually followed by some word in his own language. Thus, in case of w$gs (a wave), we find nom. plur. w$gos, but dat. plur. wegim ; so too, the dat. plur. of aiws is

2O2 THE HISTORY OF LANGUAGE. [CHAP. aiwam, while the accus. is aiwins. In Old High German the coincidence in termination between these two schemes goes further, and extends over all cases ; but since in such words as had a, o, or u, in the pre- ceding syllable umlaut had been produced in the plural by the i of the stem, only those words whose stem vowel would not admit of umlaut or modification became throughout identical with the a declension. Where the reverse was the case, the words naturally remained distinct in the plural, and a further develop- ment arose viz. that this umlaut in the plural began ; to be regarded as a sign of that number, and to be used for the purpose of marking it even in words whose etymology afforded no justification for the change, e.g. in hand, hande, which word originally belonged to the u declension. See also our remarks in Chapter V. pp. 87 and foil. 2. So far, in every case which we have discussed, we have had to do with similarity arising from pho- netic development of forms with identical functions : one or more cases of one system converged with the same cases in another system. Often, however, this same phonetic development creates a similarity between forms which were originally distinct and served dis- tinct purposes ; and we have a good instance of this in our personal pronouns, and one which is instructive as to the consequences of this phenomenon : The Gothic ik meina mis mik weis unsara uns uns izwara jus izwis izwis already shows no difference in the forms of accusative and dative plural ; but in Anglo-Saxon we find that a further stage has been reached :

XL] THE FORMATION OF NEW GROUPS. 203 In ic min m me $z2 ^in % ^e we tiser tis tis g& edwer e6w e6w we see (though separate forms for accusative still occur) that dative and accusative have become identical throughout, and so it is in the modern language with / mine me Pwethou thine thee our us ye (you) your you The double form of the nominative ye (you), and more especially the history of the pronoun for the third person, illustrate one of the consequences of such con- incidence, viz. that the language-producing community becomes accustomed to use the same form for certain sets of functions, and transfers this similarity to cases which it would not reach or, at least, has not yet reached by the aid of phonetic development alone. Let us consider first the pronoun of the third person. In Anglo-Saxon we find- Sing. Masc. Fern. Neuter. hit Nom. he hed hire his Gen. his hire him Dat. him Ace. hine hi hit. The forms which we now use for the plural are derived from a different 1 which in Anglo-Saxon gave us stem, the following plural for all three genders : Nom. $d Gen. %dra> or ^cera Dat. >&m Ace. $d 1 So, indeed, is our present nom. sing. fern. she.

204 THE HISTORY OF LANGUAGE. [CHAP. XL and here we find distinct forms for dative and accusa- tive, the latter of which has now disappeared, so that here, too (as in the case of the other personal pro- nouns), we use one form only (the original dative form) for both dative and accusative. But we have only reached this stage after a period of confusion and uncertainty, during which the historically correct form of the accusative and the new form (that of the old dative) strove for permanence. It is the very marked difference between ic (/) and me (accus.), to (thoii) and %e, we and us, which has protected the members of these pairs from becoming identical in form, notwithstanding the important fact that such a process had long since identified the nominative and accusative of all nouns and adjectives. To this influence, indeed, ye and you (both of which, when unemphatic, become yet where e is pronounced as in the before a consonant) have succumbed. Not only in this way, moreover, does such con- vergence of forms with different functions show its effect : it also causes the ordinary speaker to lose sight of such difference in function altogether. As students of Latin, and especially teachers of that language, know by sad experience, it is extremely hard for the untrained English mind to realise the function of the accusative case and the difference between this ; case and the dative may be fairly described as non- existent for the Englishman who has not learnt it from the study of other languages. This, again, influences syntax, so that a phrase like / showed him the room can be turned in the passive into The room was shown (to) him, etc., or He was shown the room, etc.

CHAPTER XII. ON THE INFLUENCE OF CHANGE IN FUNCTION ON ANALOGICAL FORMATION. THE careful consideration of such a form as / break- fasted vt'tik lead us to understand another phase in the life history of our words, and in the development of their syntactical combinations. It is well known that the word (to) breakfast is really a compound of the verb to break and the noun fast (ieiunium). Accord- ingly, we find, about the year 1400 A.D., ' Ete and be merry, why breke yee nowt your fast;' in 1653, Izaak Walton wrote, 'My purpose is to be at Hodsden before I break my fast ;' and as late as 1808, Scott writes in his Marmion, 'and knight and squire had broke their fast' l In these and similar cases, the words have retained their full and original meaning of ' to put an end to fasting by eating ; ' and the natural apprehension of this compound when employed as a noun was in the sense of the meal whereby this process is effected after the night's fasting, i.e. the first meal taken in the day. When once the verb had thus acquired the meaning of ' to take the first meal in the day,' and was next applied even in cases where so little food had been taken before that meal as to be hardly worth considering a ' meal,' the meaning of 1 Murray. Dictionary, s.v. 29 c.

206 THE HISTORY OF LANGUAGE. [CHAP. ' had been effaced by the new sense ' breaking the fast of eating the first IMPORTANT meal of the day. The change of meaning, coupled with the change in func- tion, disconnected the compound from the linguistic groups to which it had hitherto belonged, and so it came about that, after the analogy of other verbs formed from nouns, to breakfast was conjugated as a weak verb. Thus, in 1679, Everard writes, After breakfasting peaceably ; and about a century later, the word is used transitively in the sense of ' to entertain at breakfast,' e.g., They will breakfast you, or / was breakfasted^ This and all the following examples to be discussed in this chapter illustrate the point that, in the uncon- scious grouping of our words into material and modal groups, it is mainly the function of the word which causes such grouping ; and that a change of function, entailing, as it does, a change in the grouping, will often expose the word which has thus altered its meaning to the influence of analogy with other groups, though as long as it preserved its original meaning it stood quite apart from them. No doubt, however, similarity of form conduces also sometimes to this end. The group to which the word once belonged will then follow its own path of development, while the detached member will go on its new way. We have a similar instance in vouchsafe : The king vouches it SAUE (Robert of Brunne, early in fourteenth century), where we should now say : The king vouch- safes. The verb to backbite is most probably a derivative from the compound nouns back-biting (of which the earliest instance dates from 1175) and backbitter (which is found as early as 1230); while in the Early English Psalter (A.D. 1300) the past tense 1 Cf. Murray, s.v.

xii.] INFLUENCE OF CHANGE IN FUNCTION. 207 is still formed bac-bate. Gower (1393) already formed the past participle back bited^ Again, the noun brow- beating (from * to beat one's brows/ i.e. ' to lower the brows/ 'to frown'), found as early as 2 I58i, became, from a compound noun, a simple one with the mean- ing of scolding or teasing ; and gave rise to a verb to browbeat, of which the earliest known instance dates from 1603. It si > however, doubtful whether this verb has hitherto been definitely separated from the group to which etymologically speaking it belongs. The past participle brow-beat (1803 ; Jane Porter, Thad- deus) occurs, it is true, but the more usual form is as yet browbeaten. The most ordinary results of this process are, of course, all the numerous formations from nouns that have been pressed into the service of verbs /as, box, ; He boxed ; (to) dust, (to) soap, (to) dog, etc., etc. : in the case of all which, the change of function must have preceded all forms due to analogy with the groups into which the word entered solely in consequence of that change. So, again, as long as a word has an adjectival function, and even when it is used sub- stantively, but retains its original attributive meaning, it is, in English, not declined : as the POOR men ; the POOR ye have with you always ; the BLUE hats. When, then, only certain individuals belonging to the class designated by the adjective have to be indicated and not, as in the case of the poor, all the individuals possessing the quality of poverty, we resort to the addition of the word ones : as, / do not like those green hats ; I prefer the blue ONES. As soon, however, as the word loses its real signification, and passes into a proper noun, it is at once declined : as, the Grays, the 1 Murray, s.v. 2 Ibid., s.v. ; and Skeat, Etymological Dictionary, s.v.

208 THE HISTORY OF LANGUAGE. [CHAP. Pettys, the Quicklys ; the Blues, the Liberals, the 1 Conservatives, etc. It may happen that the position of the accent aids to produce change of function, as in the case of pr6- fecto (pr<5 facto), and in the very interesting case of igitur, which has been shown to be the enclitic form of agitur, originating in the common Plautine phrase (Quid agitur) Quid igitur? The case is similar with the adverbial termination ment in French and -mente in Italian, from the Latin mente. Cruellement (crudeli mente) and fierement are intelligible formations; but solidement, lourdement, etc., are formed upon their analogy. At first applied only to adverbs of manner, the termination was transferred to adverbs of time and space ; as, anciennement, large- ment. Our English termination -ly (from like) is a familiar instance of the same degradation of the final syllable: cf. godlike, by the side of godly. The word self was originally an adjective mean- ing in Anglo-Saxon and Middle-English ' the same/ and declined in apposition with the noun or personal pronoun to which it was attached to mark emphasis. It then stood in the same case, number, and gender, he selfe, his selfes, him selfum, hine selfne, etc., gen. and dat. sing. fern, hyre selfre, etc. The history of the development from this usage to our present one is not quite clear but we should remember that the ; terminations of the adjective were among the first to wear off completely, or at least to become confused and indistinct and, further, that the accusative of the ; personal pronouns, was at an early date merged into 1 Used very often in a sense quite distinct from the Liberal ones : the Conservative ones, etc. 2 King and Cookson, Principles of Sound and Inflexion, Cf. p. 285.

xii.} INFLUENCE OF CHANGE IN FUNCTION. 209 Wethe dative. thus obtain the following schematic declension : Nom. / self SINGULAR. he, she, it self my self thou self his, her, his self Gen. Ace.) mesdS thy self him, her, him self thee self Dat.j PLURAL. Nom. they selve Gen. their selve Ace.) \"iem S *ve Dat I Now, if we bear in mind that in these combinations the accent fell upon the word self (or selve), and that consequently the proclitic forms my, me, and thy, thee, in the genitive and dative had the same sound respectively, and, further, that in the feminine of the third person singular (herself}, these two cases were also alike, it does not seem strange (i) that these two cases (genitive and dative) became confused, and (2) that the word self became a noun, as exemplified in such phrases as / said it to herself. Once having changed its function, the word assumed the flection of the new group to which its new function had ittached it, and a plural form, as of a noun, arose Wheniselves, oiirselves, theirselves? once a single form served in three (genitive, dative, accusative) of the four cases, it not unnaturally succeeded in ousting L he last, and succeeded all the more easily as / self ras, of course, wrong, if self was a noun. It is not, however, an invariable rule that the new 1 This last ungrammatical form, like the singular his self (now a ilgansm), testifies to the confusion of dative and genitive.

210 THE HISTORY OF LANGUAGE. [CHAP. associations into which a word enters in consequence of its change of function entail a change of form in the word. In Latin .the word frugi was originally the dative case of a word frux, gen. frugis, meaning fruit, profit, advantage; and is actually employed by Plautus, with the full consciousness of its origin, in the phrase bonce frugi esse (Asin., III. iii. 12). In fact, this use is exactly parallel to the use of usui in bono usui estis nulli, in Plautus, Curculio, 1. 499 ; but in this case, usui, owing to its frequent occurrence, preserved the memory of its origin fresh. Cicero, however, treats frugi simply as an indeclinable adjective : Homines et satis fortes et satis plane frugi et sobrii (In Verrem, v. 27). Instances are also frequent where a change in meaning brings about a change in syntactical construction. Thus, for instance, in Latin we find that the nominative quisque is coupled with the reflective pronoun in the plural almost in the signification of singulatim? In Plautus we find prasente testibus (Amphitruo, II. ii. 203), and, in Afranius and Terence, absente nobis (Eunuchus, IV. Aiii. 7) ; in these cases the participles approach the characteristics of prepositions. similar development gave to the present participle considering its present prepositional force. Macte is used similarly. Latin is used as generally as Come ! in English, irre- spective of the number of persons addressed \\ cave^' is used in the same way. Faucis is used for ' a little in ausculta paucis (Terence, Andria, 5b36y).menHt;lafse^pei,s used in French by women equally as tSov, in Old Greek, are addressed to either one or many persons indifferently. In the same way, in late Greek, otyeXo^ and w^eXe were employed simply as conjunc- tions, without any consideration of number or person, 1 Cf. Roby, Latin Syntax, p. xxiii., and 1069, 1073.

xn.] INFLUENCE OF CHANGE IN FUNCTION. 21 I construction having been 'OXeo-0ai ewv ^original '' wcpa = ' Would that / had perished on that day ! ' In English albeit is used simply as a conjunction, and may be, in the sense of perhaps, is showing a tendency to fuse into one word, as it is actually written in American conversational language mebbe. In German we find expressions like Heb hinten uber sick das glas, 'Raise your glass high' (Uhland, Volkslieder) instead of iiber dich. In the same way we find in Latin suo loco, etc. and in Latin law ; formulae, Si sui juris sumus, where we should expect Si nostri juris stand in our own sumus ' If we (i.e. rights'). In Old Norse a middle and passive is formed by the aid of a reflective -sik (sese), which is, of course, properly applicable to the third person only : it appears later as st ; thus, at kalla, 'to call;' at kallast, 'to be called.' In the same way, we have in English the words bask and 1 where (to) (to) busk, the proper meaning of the termination has so com- pletely died out that it is possible to write busk ye. The passive is similarly formed in the Slavonic languages. Again, change of meaning influences the construc- tion in the case of numerous verbs in Latin, which are properly intransitive, but are used as transitives. Such are/^W, 2 deperire ; demori, used in the sense of ' to be mortally enamoured ' ' to marvel ' of; stupere, at ; ardere, 'to love with fire:' the last-mentioned two words approximate in sense to mirari and amare respectively, and hence the instinct of language employs them in the same government. The verb to doubt, in the etymological signification of hesitating between two beliefs, was, and is still 1 Morris, Historical Outlines, p. 6. 2 See Roby, Syntax, p. 51.

212 THE HISTORY OF LANGUAGE. [CHAP. constructed with whether. If, however, Spenser (Faery Queene) says ' That makes them doubt their wits be not their aine,' it is because the word is employed in this case, as indeed it frequently is in Shakespeare, in the sense of ' to fear.' The verb to babble, originally used intransitively, means to prattle or to chatter. When, however, it is employed in the sense of ' to speak foolish words/ or of ' to reveal by talking,' it is used with an object in the accusative case, and a passive is formed of it e.g., ; Griefs too sacred to be babbled to the world. Again, compound words, as long as they are felt as such by the speakers, are naturally treated as such; cf. the Latin word respublica, which, though we write it as a single word, was declined in both its parts, respublica, reipubliccz, etc. But, when it had once become an indivisible unit when the form rdpublique in French, or the English word republic, was formed with its various meanings, all closely resembling, but not identical with, that of the original compound, the word came to be treated after the analogy of other nouns, and the same derivatives are formed from it as from a simple form cf. republican, etc. This fact ; is, again, instanced by such forms as high-spirited +(high-spirit ed, and not high + spirited), gentleman- like (gentleman + like, not gentle + manlike), good- natured (goodnature + ed, not good + natured). Similarly, the Latin compound i (a demonstrative pronoun) + pse was at first declined as eumpse (e.g., Plautus, True., I. ii. 64), eampse, eopse, eapse, etc., all 1 When, however, which forms are found in Plautus. 1 Nay, we even find the suffix -pse attached to other parts of speech; cf. sirempse, Plaut., Amphit, Prol. 73-

XIL] INFLUENCE OF CHANGE IN FUNCTION. 213 the word came to be looked on as .a simple word, it was declined as such : ipsum, ipsam, ipso, ipsa, etc. In German there are many instances of words compounded with adverbs of place which are specially instructive as to the way in which a word may become detached from its previous use by a change of mean- ing. For instance, in modern German the usage is to say wirken AUF etwas, and not IN etwas, which was the usage even in the last century. In the same way, we speak of influence over as much as of influence on, showing that we have forgotten the significance of 1 in. The word welcome in such phrases as I made them welcome is employed as an adjective, as, indeed, it is commonly apprehended to be. It was originally a substantive, and was derived from the infinitive mood of the verb, its meaning being pleasure-comer. The word is popularly supposed to derive from wsll and come ; but the first element in the compound is really related to will the true sense being the will-comer, i.e. he who comes to please another s will. (Cf. Ger. willkommen.} The change in meaning seems due to Scandinavian influence, for in the Scandinavian lan- guages the word is really composed of the adjective well and the past participle come ; cf. Danish velkommen 2 (welcome). t The expression Quin conscendimus equos (Livy, i. 5 7) is properly Why do we not mount our horses f but is understood as Let us mount our horses ; and in accordance with such usage quin may take after it an imperative, as quin age ; or a hortative subjunctive, as quin experiamur ? The sense of cur in some cases approximates to that of quod ; and hence we find the word followed by a similar construction, in Horace, 1 See Matzner, vol. ii., p. 313, 314, etc. 2 Cf. Skeat, Etymological Dictionary, s.v.

2i4 THE HISTORY OF LANGUAGE. [CHAP. Ep. I. 8. 9 ; irasfar amicis, C2tr me funesto properent arcere veterno. The O.Fr. car underwent a similar change. Derived from qnare it meant, in the first instance, then ; as, Cumpainz Rolond, I'oliphant KAR sunez (Chanson de Roland), i.e. Compagnon Roland sonnez DONG * it next came to be used like que or loliphant ; parceque after phrases like la raison est ; and it then comes to be used with the conditional and imperative in the sense of utinam (cf. Diez, iii. 214). In O.Fr. the word par (Latin per) was used for much. It took this sense from its use in combinations like perficere, perraro, etc., but it was detached from the verb, and was habitually used in O.Fr. in such combinations as par fut proz = il fut tres preux ; and in some cases coupled with other adverbs, like moult and tant ; as, tant par fut bels = il Mail si beau, literally tant beaucoup (Chanson de Roland). The phrase survives in par trop. The Greek OVK vvv, originally not therefore, like the Latin nonne, serves to introduce a question expect- ing an affirmative answer. It then comes to be used to introduce direct positive assertions thus, OVKOVV ; tXtvOepLa ^/Aa? /xeVet ; from meaning ' Does not, then, freedom await us ' comes to mean simply ' Therefore ? freedom awaits us.' The word nanu in Sanskrit has gone through a similar development. Ne in Latin, properly the interrogative particle, comes to be used as the correlative of an : -faciatne an non faciat ; or even faciat, necne. Similarly, in Russian, the inter- rogative particle li comes to be used as the correlative of Hi (or) ; as ugodno-li vam eto ? ('Is this agreeable to you ? ; but we then get combinations like dye'laet-li, ') Hi ne dydaet (' whether he does it or no '). 1 See Cledat, Grammaire de la Vieille Langue Franchise, p. 261. 2 Cledat, p. 253.

XIL] INFLUENCE OF CHANGE IN FUNCTION. 215 The accusative with an infinitive could originally only stand in connection with a transitive verb as long as the accusative of the subject was regarded as the object of the finite verb, as audio te venire ; but the accusative and infinitive came to be regarded as a dependent sentence with the accusative as its subject, and then we find the construction after words like gaudeo, forreo (Livy, xxxiv. 4. 3), doleo (Horace, Odes, iv. 4. 62), etc., which can properly speaking take no accusative of the object connected with them as ; gaudere, dolere, infitias ire ; nay, we find it after com- binations such as spent habeo, etc. The accusative and infinitive construction then passes into sentences which depend on another accusative and infinitive, as (i) into relative sentences loosely connected e.g. ; mundum censent regi numine Deorum ex quo illud natura consequi (Cic. de Fin., iii. 19, 64) : (2) into sentences of comparison ; e.g. ut feras quasdam nulla mitescere arte sic immitem ejus viri animum esse (Livy, xxxiii. 45) : (3) into indirect questions ; e.g. quid sese inter pacatos facere, cur in Italiam non revehi (Livy, xxviii. x (4) into temporal and causal sentences ; 24) ; e.g. crimina vitanda esse, quia vitari metus non posse A(Seneca, Epist., 97. 13). similar extension of the use is found in Greek. The possessive cases mine, thine, his, her, its, our, your, their have passed into the category of adjectives, as in the case of Shall I not take mine ease in mine inn ? (i Henry IV., III. iii. 93). The instinct of language regarded mine, thine, etc., as the equivalents of of me, of thee, etc. ; and marked the function by the addition of the possessive preposition of, as in this inn of mine. Thus, again, a gerund like killing? from having the 1 See Drager, Historische Syntax, vol. ii., p. 436. 2 Cf. Mason, English Grammar, p. 64.

216 THE HISTORY OF LANGUAGE. [CHAP. xii. same form as the participle, can be used in expressions like the killing a man, instead of the killing of a man. We not only find that the word which changes its function undergoes the consequent changes in form or in syntax, but it also often happens that, owing to functional changes participated in by a certain group of words, such a group becomes detached, and thereby gains independence enough to influence other words that have cognate meanings. There are in Old English, as in German, many adverbs which are in their origin the genitives singular of strong masculine and neuter substantives, such as dceges (by day] ; but the origin of the termination has been forgotten, and the s has come to be looked upon as a merely adverbial termination. Consequently we find the adverb nihtes (by night], though niht is really feminine, and its genitive case is properly nihte. Similar formations are here- abouts inwards, othergates (Shakespeare, Twelfth > Night, V. i. 198), towards, whereabouts, etc. In the same way, the genitive plural of Anglo-Saxon substan- tives in -ung (later -ing) could be used adverbially; as, dn-ung-a, dn-ing-a, (altogether), genitive plural of an- ung, a substantive formed from an (one) : after this analogy others were formed : as, hedling, afterwards altered to headlong ; darkling, etc.

CHAPTER XIII. DISPLACEMENT IN ETYMOLOGICAL GROUPING. WE have already more than once had occasion to point out that, in our individual vocabularies, two classes of words are inextricably confused. In the first place, we employ such words and derivatives of words as we REPRODUCE by the aid of MEMORY, which recalls to us what we have frequently heard from those with whom we have intercourse. In the second place, another part of our stock of words and verbal deriva- tives is FORMED by us on the MODEL OF OTHER FORMA- TIONS of the first class. Only in a very few cases is it possible for any speaker to decide, with absolute certainty, whether any particular form which he may employ with perfect familiarity belongs to the former or the latter group. HeIf, for instance, we hear the simple sentence, ' is walking/ there is nothing which can help us to deter- mine whether the speaker is merely reproducing the word walking just as he has learnt it from others, or whether he is forming the present participle of and from the word ' walk ' after the model of other (to) similar derivatives. In the chapter on Analogy, we considered principally cases falling under the second class, in which the result of such a process as we have described proved at variance with other forms already

218 THE HISTORY OF LANGUAGE. [CHAP. existing in the language, i.e. where Analogy brought about certain changes. The cases in which the result was the mere production of what we should have reproduced by the simple aid of memory, we considered as of very small importance for the purpose of illustrat- ing the operations of Analogy. But it is far from true that they have no signifi- cance. Every time that we consciously or uncon- sciously form words ' our habit of doing by analogy,' so is strengthened, and our confidence in the results is increased and the more we enter upon domains of ; thought where we are comparative strangers, the more confidently and the more consciously do we proceed 'to make our own words.' In this process of word- making, we follow certain models in fact, we de- ; rive one form from others which exist in our own vocabulary. In words and forms reproduced by memory (though only in the case of such as these) it is, strictly speak- ing, correct to say of each form tense, person, singular or plural, or of each case that it is derived, not from what our grammars call the standard forms (such as infinitives or nominative-singulars), but from the corre- sponding older form of that tense, person, etc., in the language as it existed before. In words and forms produced, not from memory, but by analogy, i.e. by derivation according to a cer- tain model, and from words which already exist in our own vocabulary, even where our result does not differ from what we might have produced by memory, it does not at all follow that our process of derivation has been the same as that by which former speakers reached their results. For instance, suppose that there exists a class of adjectives really derived from verbs. In the course of

xiii.] DISPLACEMENT IN ETYMOLOGICAL GROUPING. 219 development of the language, these verbs approach in form to the cognate nouns, or for whatever reason- some of the verbs become obsolete. The effect will be that, in the consciousness of the ordinary speaker, the adjective appears as derived from the noun. It is our object in this chapter to study the phe- nomenon of such displacements in the etymological connections and the consequences which follow there- from. A good instance may be found in the history of the suffixes die, able, and their 1 Both these application. suffixes we owe to the French language, which, in turn, derived them from Latin. In this latter language we find the suffix bili-s, bilem, forming verbal adjectives. Where the stem of the verb ended in a consonant, the connecting vowel i was inserted : vend-e-re, vend-i-bilis. Where the stem ended in a vowel this insertion was of course unneces- sary : honora-re, honora-bilis, dele-re, delebilis, (g)no- scere no-bilis, etc. By far the greater number of these words in ble were derived from verbs in are, of which the present participle ends in ans, antem. Hence, though the words in ble were in reality not immediately derived from this participle, a feeling arose that such a connection existed. Among ' the matter- ' in French their existed numerous pairs, such groups as aimant, aimable, etc. In time, all present participles in French came to end in this termination ant, after which an adjective in able, derived from such participles, nearly always supplanted the older and correcter forms in idle, etc. Hence came forms like vendable, croyable, etc. The suffix able, introduced into English in enor- mously preponderating numbers, was there at first 1 Cf. Murray's Diet, -ble and -able.

220 THE HISTORY OF LANGUAGE. [CHAP. confined to words of French origin, but soon, by analysis of such instances as pass-able, agree-able, com- mendable, was treated as an indivisible living suffix, and freely employed to form analogous adjectives, being attached not only to verbs taken from French, but finally to native verbs as well, e.g., bearable, speakable, breakable. These verbs have often a substantive the same form, as in debat(e]-able, rat(e]-able, etc. Owing to this, a new displacement such as we are here study- ing occurred, and such words, treated as if derived FROM THE NOUN, became the models for others where able is added to nouns, such as marketable, clubbable, carriageable? salable. Another sum*, the history of which affords an instance of similar displacement is ate as verbal form- 2 ative. We find in French several past participles, some due to regular historical development of the popular language, others to deliberate adoption by the learned classes, all of which differ only from their Latin pro- totypes in having lost the termination us: e.g., confusus, 1 That ' carriageable ' is a very unusual word does not matter at all, the point is that it is formed and that it cannot be derived from a verb. 2 What follows is almost entirely taken from the article in Murray s Dictionary dealing with the suffix. Our excuse for reproducing it is the unavoidably high cost of the work, which places it beyond the reach of the ordinary student, so that a mere reference to it would be useless; and, secondly, that we believe that in Murray's other-, wise admirable treatment of the subject, one not unimportant side of the question has been overlooked. To avoid misunderstanding, we ought perhaps to assure the reader that what we give is not simply a copy of the article in question ; this will appear to any one who will take the trouble to compare the two. Our object being different, we lay more stress upon some points which are less material to I Murray ; we, however, use his facts, and wish to acknowledge our indebtedness.

XIIL] DISPLACEMENT IN ETYMOLOGICAL GROUPING. 221 Fr. confus ; contentus, content ; diversus, divers. This analogy was widely followed in later French in intro- ducing new words from Latin, and, both classes of French words (i.e. the popular survivals and the later accessions) being adopted in English provided English in its turn with analogies for adapting similar words directly from Latin by dropping the termination. This process began about 1400 A.D., and the Latin termi- nation atus gave English at, subsequently ate, e.g. desolatus, desolat, desolate. The transition of these words from adjectives and participles to verbs is ex- plained by Dr. Murray by a reference to the fact (a) That in Old English verbs had been regularly formed from adjectives : as, hwit, hwitian (' white,' ' to whiten ') ; wearm, wearmian (' warm/ * to warm/) ; etc. (6) That with the loss of the inflections, these verbs became by the fifteenth century identical in form with the adjectives, e.g., to white, to warm. (c) That, as in Latin, so in French, many verbs were formed on adjectives ; whence, again, English received many verbs identical in form with their adjectives, e.g., to clear, to humble, to manifest. These verbs, though formed immediately from par- ticipial adjectives already existing -in English, answered in form to the past participles of Latin verbs of the same meaning. It was thus natural to associate them directly with these Latin verbs, and to view them as their regular English representatives. This once done, it became the recognised method of Englishing a Latin verb, to take the past participle stem of the Latin as the present stem of the English, so that English verbs were now formed on Latin past parti- ciples by mere analogy and without intervention of a participial adjective ; e.g., fascinate, concatenate, etc. These English verbs in ate correspond generally to

222 THE HISTORY OF LANGUAGE. [CHAP. French verbs in er, e.g., separate, Fr. stparer ; this, in turn, gave a pattern for the formation of English verbs from French, e.g., isoler (Ital. isolare, Lat. insular^ Eng. isolate, etc. To this lucid and apparently adequate explanation we must, however, add another fact, which has demon- strably aided in the formation of the enormous number of English verbs in ate. From the fourteenth century onward, we find again and again such pairs as action x (1330), to act (1384) ; affliction (1303), to afflict (1393) ; adjection (1374), to adject (i43 2 ); abjection (1410), to 2 abject (1430), etc. Such pairs led to the supposition that the verbs were derivable from the nouns in tion by merely omit- ting the ion, and this was done with many nouns in ation even where another verb (itself the ground-word for that form in ation] existed by the side of it. Thus we find, e.g., aspiration (1398), to aspire (1460), the verb aspirate (1700) ; attestation (1547), to attest (1596) to attestate (1625) ; application (1493), to apply (1374), 3 to applicate (i53i). 1 The number in brackets behind these words gives the date of the earliest quotation found for their use in Murray's Dictionary. 2 It will help us to reahse the strength of the ties which united these groups, if we remember that the modern pronunciation of the ending, tion as shun is really quite modern, i.e. that, formerly, the // was in such words pronounced as tea and not as sh. The verb abject consisted therefore of the first two syllables of the noun objection, WITHOUT ANY ALTERATION. A3 carefully compiled list of all forms in ation, past participles in ate, verbs in ate, found in Diet. Murray, sub. let. A., has given the following results : Forms in ation 219. Of these the first instance belongs to the fourteenth century in n, fifteenth in 26, sixteenth in 49, seventeenth in 76, eighteenth in 23, nineteenth in 34 cases. Among the 219, the form in ation is the only one in 89 cases, distributed over the same centuries as follows, fourteenth, 2 ;

XIIL] DISPLACEMENT IN ETYMOLOGICAL GROUPING. 223 The suffix/^// forms adjectives from nouns : baleful, A.S. bealofull from bealu (woe, harm, mischief) ; shame- ful, A.S. sceamfull from sceam (shame). This ending was also added to nouns of Romance origin ; e.g., power- ful, fruitf211. In both classes, however, the word might, in very many cases, be just as well derived from a verb as from a noun, so that, e.g., thankful, which originally undoubtedly was =full of thanks, could equally well be apprehended as he who thanks ; respectful, as he who respects; etc. It is similar with such words as harmful, delightful, etc. That such a grouping has actually been made, is proved by the occurrence of such forms as wakeful, forgetful, and the dialectical urgeful ; so also the form weariful seems more likely to be inter- preted as that which wearies, than as a derivative from the adjective weary as Matzner seems to take 1 So, it. again, the form maisterful, found in Lydgate and 2 seems more likely to be taken as ' he who is Chaucer, always mastering/ than ' as he who is full of master/ which gives no sense. The suffix less, originally and still as a rule only added to nouns, could not have been used with the verb to daunt O.Fr. danter, Modern ( French, dompter, Lat. domitare, ' to tame/) if in such compounds as restless, sleepless, hopeless, useless, the noun had not been identical in form with the verb. fifteenth, 9; sixteenth, 10; seventeenth, 31; eighteenth, 15; nine- teenth, 22. There are 138 verbs in ate, 20 of which stand alone. Distribu- +tion : fourteenth century, o; fifteenth, 4; sixteenth, 53 7; seven- +teenth, 53 13; eighteenth, 13; nineteenth, i$. Of all cases where we find both the noun in ation and the verb in ate, the noun is older in 74 and the verb in 34 cases. It seems plain therefore that we may say that in English the verbs in ate are in very many cases formed from the nouns in ation, and that both are chiefly due to the sixteenth and the seventeenth centuries. 1 Vol. i., p. 433. a Goeders, p. 9.

224 THE HISTORY OF LANGUAGE. [CRAP. The history of the suffix ness, is also especially instructive for our purpose. If we go back to the oldest records of the Teutonic languages, Gothic, we find a noun, ufarassits, literally ' overness,' used in the * sense of ' abundance,' superfluity/ from ^lfar, 'over:' similarly formed was ibnassus, ' from ibns equality,' 'even/ 'equal/ This suffix assus was very frequently added to the stem of verbs which, in their turn, were derived from nouns. Thus, for instance, besides the noun lekeis (leach), we find lekinon (to cure), lekinassus (leachdom). shalks (servant), ,, shalkinon (to serve), shalkinassus (service). gudja (priest), ,, gudjinon (to be priest), gudjinassus (priesthood). frauja (Lord), fraitjinon (to rule), fraujinassus (dominion). (king), >iudanon (to be king), ftiudinassus (kingdom). In all these and similar cases, however, etymo- logical consciousness might equally well operate otherwise. It might, for instance, derive a noun meaning kingdom from another noun denoting king, or one meaning priesthood from one denoting priest. That this has been done is proved by the fact that the n has coalesced completely with the suffix assus, forming nassus, or, in its more modern form, ness. Even in Gothic, this coalescence has already been powerful enough to produce vaninassus (want) from vans (adjective = ' wanting/ 'less;' found, e.g., in wanhope = ' lack of hope/ '' wanton, = ' un- despair : educated/ 'untrained/ 'unrestricted/ 'licentious :' and wane = ' to less '). grow In Anglo-Saxon, adverbs were formed from adjectives by means of the termination e : for instance, heard, hcarde, hard ; s6^, s6^e, (' true/ cf. soothsayer (' ') and forsooth) ; wid, wide, (wide). Adjectives in lie were formed first from nouns : eor^, eor^lic, (' earth/ ' ' and earthy ') ; gdst, gastlic, (' ghost/ ghostly '), etc. ;

xiii.] DISPLACEMENT IN ETYMOLOGICAL GROUPING. 225 then, also, from other adjectives, as heard-heardlie, a&ele-afteltt, (for ar5el-lic), etc. By the side of these adjectives, we naturally find adverbs in lice, normally formed from them by the addition of e; as, ceftelice, etc. ; but as soon as, owing to phonetic decay of the terminations, the adjectives and adverbs in both sets of words (both in those with and without lie] came respectively to coincide, when, for instance, heard and hearde had both become hard, and adjectives in lie and adverbs in lice had both come to terminate in ly, then the adjective that had never ended in lie came also to be grouped with the adverb in lice, or rather ly, and ly became the special and normal adverbial termination : as in prettily, care- lessly, etc. Thus were produced a great quantity of adverbs, the adjectives corresponding to which never had the termination ly. Modern English possesses remnants of all the above original formations; as, for instance, the adverbs (with loss of adverbial e) hard, in ' to hit hard/ loud, in 'to speak loud/ etc. or, again, the adjectives ; heavenly, earthly, kingly, goodly, etc.

CHAPTER XIV. ON THE DIFFERENTIATION OF MEANING. LANGUAGE develops by the development of the vocabu- lary of individual speakers in the same linguistic community : their tendency is to produce synonymous forms and constructions in addition to those already at their disposal. Each individual is, in fact, con- stantly engaged in increasing the number of synony- mous words, forms, and constructions in the language which he speaks. One source of this superfluous development depends on analogical formation : as when in English the imperfect is assimilated to the participle, or the participle to the imperfect ; as where forms like spoke and broke appear beside spake and brake or held, beside holden. A second source of the same superfluity depend- ing on synonyms arises from the fact that of two words, each may develop its meaning on its own lines, and the meanings may come to converge so as to become one and the same. Thus, for instance, the two words, relation and relative, the former originally the abstract verbal noun, the latter an adjective, have converged in the meaning a' related ' and it person ; has thus happened that owing to this process there arise two terms for one and the same idea. To the above a third source may be added viz., the accept- ;

CH. xiv.] ON THE DIFFERENTIATION OF MEANING. 227 ance of a foreign word into a language where a native word already exists to express the same idea. Of course English is especially rich in words of this kind, owing to the large number of Norman-French words imported at the Conquest and maintained as an in- tegral part of the language ; though the process of borrowing from French has been also active since the epoch of the Conquest : such are the pairs nude, naked ; pedagogue, schoolmaster; poignant, sharp ; peccant, sin- ning ; sign, token: other familiar instances are tether, derived from the Celtic at an old date and loot, ; adopted from the Hindi, by the side of plunder. The case is, of course, similar where a synonym is adopted from another dialect, as vetch by the side of fitch, vat beside fat (a vessel), etc. But though such superfluities in language are con- tinually appearing, they have a constant tendency to disappear on the earliest possible occasion. Language is a careful housewife, who is constantly endeavouring to keep nothing on hand but what she can use, and Wecarefully to retrench the superfluous. must, of course, never suppose that any body of speakers com- bine to admit a word into the common language which they employ, and that then, finding that the word or form has its meaning already expressed in their language and is therefore unnecessary, they proceed to discard it. These new words and forms proceed in each instance from individuals, who overlook the exist- ence in their own language of a term already in use for some meaning which they need to express, and so introduce a new form : this is then employed by others, who, hearing the new form and the old, employ both alike indiscriminately. Superfluity in language, then, must be regarded as spontaneously arising, and with- out the aid of any voluntary impulse on the part

228 THE HISTORY OF LANGUAGE. [CHAP. of any individual or individuals. The language of common life is, as might be expected, most ready in freeing the vehicle of ordinary communication from superfluities, and in the differentiation of synonyms. The language of poetry and, in a less degree, of written prose, demands a store of synonyms, on which an author may draw at will, thereby forming an in- dividual style and avoiding monotony. It is as use- ful, nay, as indispensable to the poet that he should have a store of words with similar meanings which he may employ for the purposes of his artificial style, as it is for the ordinary speaker or writer to have a distinct shade of meaning attached to each of the synonyms which he employs. And as poetry makes greater demands upon the taste and powers of an author than prose, we find that the language of poetry preserves archaic forms and words which in prose have been practically obsolete. In fact such words become the stock in trade of all writers of poetry, appearing, of course, most frequently in those who seek to invest their work with a peculiarly archaic caste. Thus, the diction of Spencer must have ap- peared almost as archaic to his contemporaries as to 1 Poetry will also maintain constructions ourselves. which have a tendency in prose to become obsolete : as, meseems ; Time prove the rest. The metaphors em- ployed in old Norse poetry are very instructive on this head. They have been treated at great length in the 'Corpus Poeticum Boreale' by Vigfusson and York Powell, from whose work 2 we cite the following instances. The breast is spoken of as the mind's 1 Cf. Abbott and Seeley, English Lessons for English Readers, 2 Vol. ii., p. 446, 467, Figures and Metaphors (Kenningar) of Old Northern Poetry.

xiv.] ON THE DIFFERENTIATION OF MEANING. 229 house, memory s sanctuary, the lurking-place of thought, the shore of the mind, the bark of laughter, the hall of the heart. The eye is the moon or star of the brows, the light or levin of the forehead, the cauldron of tears, the pledge of Woden. Herrings are the arrows of the sea, the darts, the tail-barbed arrows of the deep. Ships are characterised by a host of metaphors ; as, the tree or beam, the sled, the car, the beam or timber of the sea or waves ; the steeds of the helm, oars, mast, sail, yard : and numerous other specimens of 'pars pro toto.' The most simple and obvious case of retrenchment in language is where, out of several similar forms and phrases, all disappear and are disused except a single one as where to grow is used instead of to wax; to go, ; Weinstead of to fare, etc. must look upon these retrenchments in language as mainly due to indi- viduals each speaker expresses himself more or less ; unconsciously with a certain consistency, and uses, generally speaking, what we may properly call his own dialect. It is owing to such individual influence that the distinctions in language which we call dialects arise, and thus the different opportunities for choice form a main source of the distinctions of dialect. In addition to this negative process of simply dropping what is useless, there is the positive process of utilising what is superfluous in language by differ- entiation of meaning in the case of synonymous words and phrases. This process is no more the result of conscious purpose than the other. Since each indi- vidual has gradually to learn the different senses of words, inflections, particles, etc., it is clear that when there are several synonyms in use each of which has several shades of signification he will almost certainly hear one of them used in one, and another in another of these meanings. If, for instance, we represent the

230 THE HISTORY OF LANGUAGE. [CHAP. full meaning of a word in its different shades by the A B Cletters + ++ D, and, similarly, that of its synonym by +a -f b c 4- d, the probability almost amounts to certainty that when a learner first hears the former word, the shade of meaning (say B] in which it happens to be employed will differ from that (say cC] in which he first learns the use of -the latter. He will then inevitably, though perhaps unconsciously, attach by preference these particular shades of mean- ing to the two words and will continue to do so, ; unless stronger impulses, such as frequent use in other meanings by surrounding speakers, force him to dis- card the differentiation which he has established. But from the moment when he begins to use, and as long as he uses the word consistently in one sense, he will influence others in the same linguistic community, and lay the basis for definite acceptance of the word in a particular or special sense. Nor, again, must we assume that a differentiation in sound was purposely and consciously made by speakers with a view to differentiate meanings. Cases taken from modern languages may serve to show the unreasonableness of such assumptions. Especial atten- tion has been paid by writers on Romance Philology to the ' occurring in their own languages. ' doublets By ' doublets ' we mean the double derivative forms of one and the same word (such as raison, l and reason/ ration, ' both coming from rationem) : forms allowance/ commonly appearing in a language at two different periods in the history of the language, and invested, in spite of their common origin, with distinct and special senses. The name of ' doublets ' was first applied to them by Nicolas Catherinot, who, as early as 1683, published a list of those which he had ob- served in French, but without giving the reasons for

xiv.] ON THE DIFFERENTIATION OF MEANING. 231 the phenomenon. How imperfect the philological knowledge of his day was may be seen from the following specimens of ' doublets ' which he gives : from BATTUERE, Low Latin for ' to fight/ he derived both battre (to fight) and tuer (to kill) : from GRAVLS (heavy), grave, serious brave, brave : from MARMOR ; (marble), marble, marble marmot, 1 A. ; guinea-pig. Brachet has collected many other specimens in the work cited below : Coelho has made a collection from the Portuguese in the Romania, II. 281, 2 sqq. It must, however, be noticed that many of the doublets cited in these works stand outside of the class of those with which we have to deal, and such cannot be taken as real cases of differentiation. For instance, a loan word may immediately upon its intro- duction have been accepted in a sense different from that borne by the word of the same origin which already existed in the language : as in the case of chante'e (sung, fern. past, part.) and cantata (cantata, a piece which is sung, as distinguished from a sonata, a piece which is sounded or played), borrowed from the Italian by the French of sexte (term in music and ; ' the sixth book with its doublet sieste (the hour of ') rest) borrowed from the Spanish siesta, both derived from the Latin sextam; of facon (manner) with its doublet fashion, borrowed from the English, both from Latin factionem, 'a making.' Thus, again, the French 1 See Brachet, Dictionnaire des Doublets, Appendice. Paris, 1868. 2 Other works on doublets are Romanische wortschopfung, by Caroline Michaelis, Leipzig, 1876. Latin doublets, by M. Breal, in the Memoires de la Soriete de Linguistique de Paris, i. 162, sqq. (1868). For German, O. Behagel, Die Neuhochdeutschen Zwillingsworter, Germania, 23, 257, sqq. For English doublets, cf. Matzner, Englische Grammatik, i. 221; and Skeat, Principles of English Etymology, p. 417 ; besides the appendix to his Lexicon.

232 THE HISTORY OF LANGUAGE. [CHAP. chose (a thing) and cause (a cause) alike owe their origin to the Latin causam, but the meanings were not differentiated in France : cause was borrowed as a law- term long after chose had developed into the general meaning of thing. It is the same, moreover, with such English doublets as ticket, etiquette: army, armada : orison, oration : penance, penitence. Such doublets as these, and guitar, zither, cithara may be called pseudo-doublets, producing as they do the effect of differentiation, but serving really as labels to designate a foreign idea or object. Nor, again, must we include cases in which a word became grammatically isolated and then received a special meaning ; such as where ' in German, is now employed with besduoden,' the signification of ' while ' besclwden ' is used modest,' as the true participial form, and never means, or has meant, ' modest.' Similarly, in French, we have savant (a scholar) originally used as synonymous with present participle sachant (knowing) but in modern French as an adjective or noun only, whilst sachant has always remained present participle and no more : amant, the present participle of amare (to love) is used as a substantive 1 only. There are, however, other cases in which words are really differentiated ; that is to say, cases in which two words, whose meaning we know to have been identical, have come to be accepted in different meanings. This is a genuine process of economy in language. In French sattaquer a and s attacker a at one time were used with identically the same meaning and employed indifferently. Attaquer is used in the sense of ' in this line of the four- * attacker teenth century Une riche escarboucle le mantel ataqua =a rich carbuncle attached held) the mantel (' ( ') 1 See Matzner, Fr. Gr., p. 223.